the oven a t 110” C. until immediately before ube. 5-mv. recorder coupled to an Instron Integrator is used for digital readout of peaks. Procedure. Ai mixture of 1 ml. of sample, 1.5 ml. of hexamethyldisilazane, 0.5 gram of sand, and 0.5 gram of Drierite is refluxed for 45 minutes on a 200” C . sandbath, cooled, and a 0.1-ml. sample injected into the chromatograph. Peak areas are integrated and percentages of individual peaks evaluated as fractions of the total integrand, ignoring area factors. A \
RESULTS AND DISCUSSION
measured percentages. The peaks exhibit very little tailing and are all well separated. The hindered phenols (having o,o’-methyl substituents) are present in too small a proportion to demonstrate completeness of reaction. This is dealt with along with theoretical considerations in our previous paper (4). ACKNOWLEDGMENT
The alkylphenols were furnished by Donald C. Jones of the Research Division of Consolidation Coal Co., who also prepared the 2,4-xylenol phosphate. LITERATURE CITED
(1) Averill, W., 2nd Int. Symposium of
Figure 1 is a chromatogram of an accurately proportioned mixture of pure phenols listed in Table I. Areas listed are the digital areas multiplied by attenuation factors. Good agreement is obtained between known and
Gas Chromatog., East Lansing, Mich.,
1-7 fl961). ( 2 ) Bergmann, G., Jentasch, D., Angew.
Chem. 70, 192 (1958). (3) Brooks. V. T., Chem. & Znd. (London) 1959, i3i7. ( 4 ) Freedman, R. W., Croitoru, P. P., ANAL.CHEM.36, 1389 (1964).
(5) Friedman, S., Kaufman, M . L., Wender, I., J . Org. Chem. 27, 764-5 (1962). (6) Friedman, S., Steiner, W. A., Wender, I., Fuel 40, 33-45 (1961). ( 7 ) Friedman, S, Zahn, L., Kaufman, M. L., Wender, I., Bur. Mi.zes Bulletin 609, (1963). (8) Grant, D. W., T ’ a u g e , G A , “Gas Chromatography, 1961, M. van Swaay, ed.. pp. 305-14, Butterworths, London, 1962. (9) Langer, S. H., Pentages, P., Wender, I., Chem. & Ind. (London) 50, 1664-6 (1958). (10) Paterson, A. R., “Gas Chromatography,” (2nd Int. Symposium, Analysis Instrumentation Ilivision of the Instrument SOC. of America, June 1959), H. J. Noebels, ed., pp. 223-6, Academic Press, Sew York, 1961. (11) Sassenberg, W., Wrabeta, K , 2. A n d . Chem. 184, 423-7 (1961). ROBERTW. FREEDMAN GEORGE 0. CHARLIER Consolidation Coal Co. Library, Pa.
Adsorption of Cobalt( 111) Trisethylenediamine at Mercury Electrodes SIR: In a recent publication, Anson ( 1 ) reported that he was unable to
detect adsorption of C ~ e n at ~ +8, hang~ ing drop mercury electrode using the potentiostatic current integration method (2, S), whereas we (4) had reported detection of a small surface excess by the chronopotentiometric method. He concluded that adsorption of this ion is absent, and attributed our finite intercepts in i r us. l/i plots to double layer charging effects. The following statements are made in an effort to clarify, in so far as possible, this apparent disagreement. I n the measurement of chronopotentiometric transition times, an effort is made to eliminate double layer charging from each transition time measurement. Admittedly, this is not necessarily an exact procedure because of possible errors in the graphical methods used in evahating transition times. However, we satisfied ourselves of the adequacy of the method we used by determining a set of transition times for C d + 2in the same range of i and 7 as for C ~ e n and ~ + ~finding that the i7 us. l l i plots indeed had negligible intercepts. It should be noted that negligible intercepts were a l ~ o observed for C0en3+~under certaiin conditions, notably in the absence of C1- which should enhance adsorption of the cobalt species by ion pair formation with adsorbed chloride. The potentiostatic current integration method, on the other hand, makes no provision for removing the double layer charging current from each individual measurement. Instead, the intercept of the Q us. t 1 I 2 plots includes
the sum of current required for double layer charging plus reduction of the adsorbed solute. Therefore, any change in double layer charging caused by the adsorption of reactant cannot be taken into account in running a blank. As a result, the sensitivity of the method is considerably decreased and the erroneous conclusion that adsorption is absent may be reached in cases in which adsorption occurs to a limited extent only, particularly if adsorption leads to a decrease in double layer capacity. However, increases as well as decreases can occur, particularly with charged adsorbates. Independent evidence exists for adsorption of Coen3f3 ion. The electrocapillary curve for an equimolar mixture of Coen3C12 and Coen3Cla in 1.lf NaC104 and O.1M en shows a shape typical of cation adsorption when compared with the supporting electrolyte which contains a concentration of chloride ion (added as NaCl) equivalent to that added with the cobalt cations. The deviation between the two curves begins a t -0.10 to -0.15 volt (US. S.C.E.) and increases with increasing negative potential. Because of the reversible behavior of the couple, the open circuit potential is a function of the ratio of concentrations of the two forms of cobalt complex in solution. With C ~ e n alone, ~ + ~ the open circuit potential was about -0.18 volt (4) at which point the electrocapillary curve shows only a small amount of adsorption. This may be the reason that Anson failed to detect adsorption. The unexpected result in our work was that in an equimolar mixture of Coen3+3and
C ~ e n ~a +smaller ~, amount of adsorption of C ~ e n was ~ + observed ~ than in a solution containing only the osidized form, Two opposing effects can be visualized. Because of the more negative open circuit potential, increased adsorption of cations would be expected. On the other hand, primary adsorption of chloride, together with ion pair formation between chloride and the cobalt (111) trisethylenediamine cation, appears to play an important role. The primary adsorption of chloride would be decreased by the more negative open circuit potential. In any case, the electrocapillary curves indicate a greater amount of total adsorption a t more negative potentials. I t would appear, both from the chronopotentiometric observations and the impedance behavior discussed below that the adsorption of cobalt(I1) is greatly favored over that of cobalt(II1) from an equimolar mixture. Unfortunately, the adsorption of cobalt(I1) could not be studied because of kinetic complications in the oxidative electrode process ( 4 ) . I n brief, the electrocapillary behavior appears to give unequivocal evidence for adsorption of the cobalt trisethylenediamine comple\- but it does not distinguish between the relative amounts of cobalt(II1) and cobalt(I1) adsorbed a t any potential. The anomalous faradaic impedance behavior of the C0rn3+~-Coen,+z couple can be explained by assuming additional electron exchange between adiorbed reactants and electrode ( 6 ) . The anomaly in faradaic impedance could be represented by a large capacitance in series with a small resistance, the com VOL. 36, NO. 9 , AUGUST 1964
1881
bination in parallel with the double layer capacitance. The magnitude of the correction, as represented by the size of the anomalous capacitance, increased with increasing cobalt concentration [always a t 1 : 1 mole ratio of cobalt(I1) to cobalt(III)] up to 2.17 mM. Beyond that concentration, the anomaly could be represented by a pure capacitance in parallel with the double layer capacitance, and the magnitude of the added capacitance decreased from 161 pf. per sq. cm. to 24 pf. per sq. cm. as the concentration of cobalt was increased from 2.17 to 8.25 mM. This effect was attributed to a change in ratio of the adsorbed reactants [presumably in favor of Co(I1) to permit minimization of charge density] with
increasing solution concentration. While it is true that a more detailed knowledge of the impedance behavior would be desirable, particularly with respect to the change in double layer capacitance upon adsorption of reactants, no satisfactory alternative explanation to adsorption has been advanced for the anomalous faradaic admittance. Of the methods under consideration, the impedance method appears to be the most sensitive to adsorption of reactants, but the least straightforward in interpretation. Conversely, the current integration method appears to be the least sensitive, but the most adaptable to a straightforward evaluation of surface excess concentration.
LITERATURE CITED
( 1 ) Anson, F. C., AXAL. CHEM.36, 932 (1964). ( 2 ) Zbid., p. 520.
( 3 ) Christie, J., Lauer, G., Osteryoung, R. A,, Anson, F. C., Ibid., 35, 1979 (1963). ( 4 ) Laitinen, H. A , , Chambers, L. M., Ibid., 36, 5 (1964). ( 5 ) Laitinen, H. A., Randles, J. E. B., T r a n s . Faraday SOC.51, 54 (1955).
H. A. LAITINEN Soyes Chemical Laboratory University of Illinois Urbana, Ill. L. M . CHAMBERS Ivorydale Technical Center Procter and Gamble Co. Cincinnati) Ohio
A Simple Method for the Determination of Hydrogen in the Presence of Oxygen and/or Nitrogen SIR: NPO has recently been used in radiation and photochemistry as a specific scavenger for solvated electrons ( 1 ) to yield nitrogen gas. In these and other systems, there is a need for an accurate yet simple analytical method for gaseous mixtures of HP,Nz, and/or 02. The conventional combustion method is not accurate for determinations in mixtures containing only a few per cent of one component in the presence of an excess of the others. Kuhn ( 2 ) investigated the use of silica gel for the selective adsorption of components of gas mixtures at low temperature. Shinohara, Shida, and Saito (3)used this method a t low total pressures for the analysis of mixtures containing mainly nitrogen with a few per cent of hydrogen. Using silica gel in a trap cooled by liquid nitrogen, they claimed that nitrogen only was adsorbed, but not hydrogen. The results of Kuhn ( 2 ) at relatively high pressures indicated some adsorption of hydrogen and incomplete separation. Indeed, using the same conditions as Shinohara and various samples of silica gel, some hydrogen always remained adsorbed on the silica gel, For example: Table I.
,
10
8
16
24 HYDROGEN
32 p PRESSURE
Figure 1. Percentage of H? adsorbed on silica gel as a function of Hz pressure, at -90" K.
using Davidson PA 100 Silica Gel, between 0 and 25 microns partial pressure of hydrogen, the percentage of hydrogen thus adsorbed was independent of hydrogen pressure and of the pressure of N2 and/or O2 present. Above 25 microns pressure, the percentage increased as shown in Figure 1. In all these experiments, the silica gel was far from being saturated with N2 or 02.
Determination of Hydrogen in Presence of Nitrogen and/or Oxygen
Sumbers refer to pressure in microns in a total volume of 300 ~ m . ~ Cornpn. calcd. Compn. found, from predetd. 7c Original rompn. pumping method adsorption H2 Sample H2 S2 and/or 0 2 HI S2 and/or O2 HI Np and/or O 2 3.3 22.2 3.4 22.3 3 3 H2 + 0 2 22 3 H?
+ + N2 0 2
H?+ ?;2
I& HZ, S2, 0 I
